ESTIMATING VEHICLE OPERATING COSTS DUE TO PAVEMENT SURFACE CONDITIONS

Transcription

1 0 0 ESTIMATING VEHICLE OPERATING COSTS DUE TO PAVEMENT SURFACE CONDITIONS IMEN ZAABAR, Ph.D. Visiting Research Associate, Department of Civil and Environmental Engineering, Michigan State University, S. Shaw Lane, Room, East Lansing, MI -, USA. zaabarim@egr.msu.edu KARIM CHATTI, Ph.D. (Corresponding Author) Professor and Acting Associate Dean for Research College of Engineering Michigan State University Engineering Building, S. Shaw Lane,East Lansing, MI - Ph: --, Fax: -- chatti@egr.msu.edu Paper submitted for presentation in the rd Annual Meeting of Transportation Research Board and publication in the Journal of the Transportation Research Board November, 0 Text =, Figures = x 0 =,000 Tables = x 0 =,00 Total =, TRB 0 Annual Meeting

2 0 ABSTRACT This paper presents a summary of findings on the effect of pavement conditions on vehicle operating costs. Specifically, the paper provides estimates on fuel consumption (FC), repair and maintenance (R&M) and tire wear (TW) costs with increasing pavement roughness (IRI) and texture (MPD) across all classes of vehicles and types of pavements investigated. The most important cost components affected by roughness are fuel consumption followed by repair and maintenance, then tire wear. An increase in IRI of m/km (. in/mile) increases fuel consumption of passenger cars by % to % irrespective of speed. For heavy trucks, this increase is % to % at highway speed ( km/h or 0 mph) and % to % at low speed ( km/h or mph). Surface texture (MPD) and pavement type have no effect on fuel consumption for all vehicle classes except for heavy trucks. An increase in MPD of mm (0.0 in.) increases fuel consumption by.% at km/h ( mph) and % at km/h ( mph). The effect of pavement type on fuel consumption was statistically not significant for all light vehicles, and only statistically significant for heavy trucks at low speed in summer conditions. Heavy trucks driven on AC pavements consume % more fuel than on PCC pavements at km/h ( mph) at 0 o C. No data was available for heavy trucks in winter. For repair and maintenance, there is no effect of roughness up to IRI of m/km (0 in/mile). Beyond this range, an increase in IRI up to m/km ( in/mile) increases repair and maintenance cost by % for passenger cars and heavy trucks. At IRI of m/km ( in/mile), this increase is up to 0% for passenger cars and 0% for heavy trucks. An increase in IRI of m/km (. in/mile) increases tire wear of passenger cars and heavy trucks by % at km/h ( mph). TRB 0 Annual Meeting

3 Zaabar, I. and Chatti, K. 0 0 INTRODUCTION The objective of the research performed under NCHRP Project - was to recommend models for estimating the effect of pavement conditions on Vehicle Operating Costs (VOC). These effects are essential to sound planning and management of highway investments, especially under increasing infrastructure demands and declining budget resources. The recommended models reflect current vehicle technologies in the United States. The research focused only on the cost components that are mostly affected by pavement conditions, namely, fuel consumption, repair and maintenance costs and tire wear. The research does not include the effect of pavement conditions on changes in travel time, nor does it consider the safety-related or other implications of pavement conditions. The paper begins with a summary of findings from the NCHRP - project on the effect of pavement conditions on fuel consumption (FC), repair and maintenance (R&M) and tire wear (TW) costs. The details of the research were reported in the NCHRP report 0 (). This is followed by case studies and results from the application of the mechanistic-empirical models recommended in this research. First, a large amount of data and information was collected, reviewed and analyzed to identify the most relevant VOC models. The review was focused on research that has identified pavement surface conditions that affect VOC costs. Next, a large field investigation involving surveys to collect pavement condition data and field trials to collect fuel consumption and tire wear data were conducted. These data were used to calibrate and validate the Highway Development and Management () fuel consumption and tire wear models (HDM ) and estimate the effect of pavement conditions on these components. The research also involved the collection of repair and maintenance data of vehicle fleets from two DOTs (Michigan and Texas). The fleet data were used to develop R&M models that consider the paved surfaces conditions encountered in the United States and address the full range of vehicle types. TABLE summarizes the unit costs used in this study. TABLE HDM Default Representative Vehicle Classes - Unit Costs Vehicle class Fuel Cost ($/gallon) Fuel Cost ($/liter) Tire Cost ($/tire) Repair and maintenance costs ($/mile )* Repair and maintenance costs ($/km)* Small car $. $0. $ Medium car $. $0. $ Large car $. $0. $ Van $. $0. $ Four wheel drive $. $0. $ Light truck $. $0. $ Medium truck $. $0. $ Heavy truck $. $.0 $ Articulated truck $. $.0 $ Mini bus $. $0. $ Light bus $. $0. $ Medium bus $. $.0 $ Heavy bus $. $.0 $ Coach $. $.0 $ *These costs are repair and maintenance costs caused by roughness only and are estimated based on data from 00. These costs are estimates for 0. TRB 0 Annual Meeting

4 Zaabar, I. and Chatti, K RESULTS Fuel Consumption (FC) The effect of pavement conditions was investigated using five instrumented vehicles to measure fuel consumption (FC) over different pavement sections with different pavement conditions. This data was used to calibrate the HDM FC model. The calibrated models were verified and were found to adequately predict the fuel consumption of five different vehicle classes under different operating, weather, and pavement conditions (,). The calibrated HDM fuel consumption model is listed in TABLE. The analysis showed that the HDM fuel consumption model, after appropriate calibration, adequately predicted the fuel consumption of five different vehicle classes under different operating, weather, and pavement conditions. Also, because the key characteristics of representative vehicles used in the current HDM model vary substantially from those used in the U.S., the current model (i.e., without calibration) predicted lower fuel consumption than actually consumed. TABLE and FIGURE present the predictions by the calibrated HDM FC model in terms of consumption and costs, respectively. The increase in fuel consumption was computed from the baseline IRI of m/km (. in/mile). The table and the figure were generated at a representative average temperature of C (. F) with a mean profile depth of mm (0.0 in) and a grade of 0%. The table shows that an increase in IRI of m/km (. in/mile) increases fuel consumption of passenger cars by % to % irrespective of speed. For heavy trucks, this increase is % to % at highway speed ( km/h or 0 mph) and % to % at low speed ( km/h or mph). The analysis assumed that there is no interaction between the effect of roughness (unevenness) and surface texture given that their wavelength ranges are independent. The model showed that pavement surface texture has an effect on fuel consumption only for heavier trucks. For example, a mm (0.0 in.) decrease in mean profile depth (MPD) will result in a decrease in fuel consumption of.% and.% at and km/h ( and mph) speeds, respectively. The analysis showed that there is no interaction between the effect of roughness and pavement type. The analysis also showed that the difference in fuel consumption between asphalt and concrete pavements could only be detected at low speed and for heavy and fully loaded light trucks in summer conditions (). Heavy trucks driven over AC pavements will consume about % more fuel than over PCC pavement at km/h in summer conditions. The effect of pavement type was statistically not significant at higher speeds. No data was available for heavy trucks in winter. Reduction in vehicle fuel consumption is one of the main benefits that should be considered in technical and economic evaluations of road improvements considering its significance. This research showed that a decrease in IRI by m/km (. in/mile) will result in a to % decrease in the fuel consumption for passenger cars. This would save about to billion gallons of fuel per year of the 00 billion gallons consumed annually by the million vehicles in the U.S. With today s gas prices, this will translate to about to billion dollars. TRB 0 Annual Meeting

5 Zaabar, I. and Chatti, K. TABLE Calibrated HDM fuel consumption model () Name Description Unit ( ) 00 * max, * * υ Fuel Consumption (FC) FC = ( α ξ Ptot ( + dfuel )) ml/km υ Vehicle Speed m/s α Fuel consumption at Idling ml/s Engine efficiency (ξ) = ξ b + ehp ( P P eng ) tot P max ml/kw/s ξ b Engine efficiency ml/kw/s Pmax Rated engine power kw ehp engine horsepower hp dfuel Excess fuel due to congestion as a ratio (default = 0) dimensionless Ptr Ptot = + Paccs + Peng for Ptr 0, uphill/level Total power (P tot ) edt kw P = edtp + P + P for P < 0, downhill tot tr accs eng tr edt Drive-train efficiency factor dimensionless Power required to overcome internal engine friction (0 percent of P eng kw the engine and accessories power) Pengaccs = KPea * Pmax* Engine and accessories power (P engaccs = P eng + P accs ) ( Paccs a + Paccs a Paccs a ) RPM RMPIdle _ ( _ 0 _ * RPM0 RPMIdle kw KPea Calibration factor dimensionless b+ b * a * c Paccs _ a= * a 0 PctPeng Paccs_a a = ξ b * ehp * kpea * P max* 0 factor b = ξ b * kpea * P max c = α Paccs_a0 Ratio of engine and accessories drag to rated engine power when dimensionless traveling at 0 km/h ( mph) PctPeng Percentage of the engine and accessories power used by the engine % (Default = 0%) R P M = a 0 + a * SP + a * SP + a * SP Engine speed (RPM) SP = m ax( 0, v) Rev/min a0 to a Model parameter RPM0 Engine speed at 0Km/h ( mph) Rev/min RPMIdle Idle engine speed Rev/min ν Traction power ( P ( Fa+ Fg+ Fc+ Fr+ Fi) tr ) Ptr = 00 kw TRB 0 Annual Meeting

6 Zaabar, I. and Chatti, K. TABLE (Continued) Aerodynamic forces (Fa) Name Description Unit Fa = 0. * ρ * C D * AF * υ N Mass density of the air (ρ) ( ). ρ = 0.0+.*.* ALT 0.00* Tair Kg/m ALT The altitude above sea level (Default = 00 m) m Tair Temperature of the air (Default = o C) o C CD Drag Coefficient dimensionless AF Frontal Area m υ Vehicle Speed m/s Gradient forces (Fg) Fg = M * GR* g N M Vehicle weight Kg GR Gradient radians g Gravity (Default =.) m/s Curvature forces (Fc) M * υ M * g * e R F c = m a x ( 0, * 0 ) N w * C s N R curvature radius (Default = 000) m Superelevation (e) e = max( 0,0. 0.* Ln( R) ) m/m Nw Number of wheels dimensionless Tire stiffness (Cs) M M C s = a 0 + a * + a * N w N w kn/rad a0 to a Model parameter Rolling resistance (F r ) Fr CR *( b* Nw CR* ( b * M b* υ )) = + + N CR Rolling resistance tire factor factor Rolling resistance parameters (b, b, b) b= * WD b = 0.0 / W D b = 0.0 * Nw / W D parameters WD Wheel diameter m CR= Kcr a0+ a* Tdsp+ a* IRI + a* DEF factor CR [ ] Kcr Calibration factor factor a0 to a Model coefficient dimensionless Tdsp Texture depth using sand patch method Tdsp=.0* MPD+ 0. mm MPD Mean Profile Depth mm IRI International roughness index m/km Deflection (DEF) DEF = ( Tair / 0) * ( * e 0. 0* υ ) mm Inertial forces (F i ) Fi = M * ( a0+ a* arctan( a / υ )) * acc N acc Vehicle acceleration m/s a0 to a Model parameter dimensionless TRB 0 Annual Meeting

7 Zaabar, I. and Chatti, K. 0 TABLE Effect of Roughness on Fuel Consumption Speed Or Or Or 0 Vehicle Class Base (ml/km) Adjustment factors from the base value* Calibrated HDM model Base (mpg) Adjustment factors from the base value* Medium car Van SUV Light truck Articulated truck Medium car Van SUV Light truck Articulated truck Medium car Van SUV Light truck Articulated truck m/km =. in/mile * The table was generated at a representative average temperature of C (. F) with a mean profile depth of mm (0.0 in) and a grade of 0%. Tire Wear (TW) The effect of pavement conditions on tire wear was investigated in this research using field data for a passenger car conducted in the NCHRP - project and truck tire wear data collected from the National Center for Asphalt Technology (NCAT). The HDM model was calibrated to adequately predict tire wear of passenger cars and articulated trucks. The calibrated HDM tire consumption model is listed in TABLE. TABLE and FIGURE present tire wear as a function of IRI for all vehicle classes at, and km/h (, and 0 mph) in terms of consumption and costs, respectively. The table and figure were generated at C (. F) when the MPD is mm (0.0 in) and grade is 0%. These data show, for the same IRI value, that tire wear increases with increasing speed, and that the roughness effect is higher at higher speeds. As an example, the results show that a decrease in IRI by m/km (. in/mile) will result in about % decrease in tire wear for passenger cars. Assuming that the average annual mileage for a passenger car is 000 km (,000 miles) and the average price of a tire is $0, the million vehicles will consume about. billion dollars per year. Therefore, a decrease in IRI by m/km (. in/mile) will save. million dollars per year. TRB 0 Annual Meeting

8 Zaabar, I. and Chatti, K. Fuel cost (cents/mile) Fuel cost (cents/mile) 0 0 Passenger car SUV Van (a) Light vehicles at km/h ( mph) Articulated truck Heavy truck (b) Trucks at km/h ( mph) Fuel cost (cents/mile) 0 Fuel cost (cents/mile) Fuel cost (cents/mile) Passenger car SUV Van (c) Light vehicles at km/h ( mph) 0 0 Fuel cost (cents/mile) 0 Articulated truck Heavy truck (d) Trucks at km/h ( mph) 0 Passenger car SUV Van Articulated truck Heavy truck (e) Light vehicles at km/h (0 mph) (f) Trucks at km/h (0 mph) Passenger car SUV Van Articulated truck Heavy truck FIGURE Effect of Roughness on Fuel Costs TRB 0 Annual Meeting

9 Zaabar, I. and Chatti, K. TABLE Calibrated HDM Tire Consumption Model () Name Description Unit Number of equivalent new TW T EQN T = tires (EQNT) VOL % new tire /km VOL Tire volume dm Total change in tread wear TWT = C0tc+ Ctcte TE (TWT) dm/00 km C0tc The tread wear rate constant dm/00 km Ctcte The tread wear coefficient dm/mnm The tire energy (TE) CFT + LFT TE= NFT MNm/00 km The circumferential force on ( + CTCON * dfuel) *( Fa+ Fr+ Fg) CFT = the tire (CFT) NW N CTCON The incremental change of tire consumption related to congestion ratio dfuel The incremental change of fuel consumption related to congestion ratio Fa The aerodynamic forces (TABLE ) N Fr The rolling resistance forces (TABLE ) N Fg The gradient forces (TABLE ) N The lateral force on the tire Fc LFT = (LFT) NW N Fc The curvature forces (TABLE ) N Nw Number of wheels dimensionless The normal force on the tire M * g NFT = (NFT) NW N M Vehicle mass kg G Gravity (Default =.) m/sec TABLE Effect of Roughness on Tire Wear Rates Speed Or Or Vehicle Class (number of wheels) Baseline conditions (%/km) Baseline conditions (%/mile) Adjustment factors from baseline conditions Medium car () Van () SUV () Light truck () Arti. truck () Medium car () Van () SUV () Light truck () Arti. truck () Medium car () Van () SUV () Light truck () Arti. truck () Or 0 % of the volume of a new tire m/km =. in/mile TRB 0 Annual Meeting

10 Zaabar, I. and Chatti, K. Tire cost (cents/mile) Tire cost (cents/mile) Passenger car SUV Van (a) Light vehicles at km/h ( mph) Articulated truck Heavy truck (b) Trucks at km/h ( mph) Tire cost (cents/mile) Tire cost (cents/mile) Passenger car SUV Van (c) Light vehicles at km/h ( mph) Articulated truck Heavy truck (d) Trucks at km/h ( mph). Tire cost (cents/mile) Tire cost (cents/mile) Passenger car SUV Van (e) Light vehicles at km/h (0 mph) Articulated truck Heavy truck (f) Trucks at km/h (0 mph) Passenger car SUV Van Articulated truck Heavy truck FIGURE Effect of Roughness on Tire Costs TRB 0 Annual Meeting

11 Zaabar, I. and Chatti, K. 0 Repair and Maintenance Costs (R&M) Two different approaches for estimating repair and maintenance costs induced by pavement roughness were developed and used in this paper: () An empirical approach that introduced adjustment factors to update the tables reported in the Texas Research and Development Foundation (TRDF) study by Zaniewski et al. (), and () a mechanistic-empirical (M-E) approach that involves fatigue damage analysis using numerical modeling of vehicle vibration response developed by Zaabar et al. (). The results from the M-E approach were compared to the empirical results, and were found to be very close up to an IRI of m/km or in/mile (typical IRI range in the U.S), with a standard error of about % (). TABLE and FIGURE summarize the repair and maintenance costs per km (mile) as a function of IRI for all vehicle classes and grade of 0%. The results show that there is no effect of roughness up to IRI of m/km (0 in/mile). Beyond this range, an increase in IRI up to m/km ( in/mile) will increase repair and maintenance cost by % for passenger cars and heavy trucks. At IRI of m/km ( in/mile), this increase is up to 0% for passenger cars and 0% for heavy trucks. TABLE Effect of Roughness on Repair and Maintenance Costs Speed Or Vehicle Class Average repair and maintenance cost * ($/km) Average repair and maintenance cost * ($/mile) Baseline conditions ($/km) Baseline conditions ($/mile) Adjustment factors from baseline conditions Medium car Van SUV Light truck Arti. truck Medium car Van Or SUV Light truck Arti. truck Same as above Medium car Van Or SUV Light truck Arti. truck m/km =. in/mile *these costs were only repair costs related to damage from vibrations and estimated using the data collected from Michigan and Texas DOT. As an example, if we consider the IRI distribution of the U.S. road network (FIGURE ), about % of the roads have an IRI higher than m/km (0 in/mile). Assuming that the average annual mileage for a passenger car is 000 km (,000 miles), and a total of million cars travel on the US road network, the repair and maintenance cost for passenger cars in the U.S. can be estimated to be anywhere between and billion dollars per year (corresponding to vehicle speed ranging from to km/h, or to 0 mph). TRB 0 Annual Meeting

12 Zaabar, I. and Chatti, K. Repair and maintenance cost (cents/mile) 0 Repair and maintenance cost (cents/mile) 0 Passenger car SUV Van Articulated truck Heavy truck Repair and maintenance cost (cents/mile) (a) Light vehicles at km/h ( mph) Repair and maintenance cost (cents/mile) 0 (b) Trucks at km/h ( mph) 0 Passenger car SUV Van Articulated truck Heavy truck (c) Light vehicles at km/h ( mph) (d) Trucks at km/h ( mph) Repair and maintenance cost (cents/mile) 0 Repair and maintenance cost (cents/mile) 0 0 Passenger car SUV Van Articulated truck Heavy truck (e) Light vehicles at km/h (0 mph) (f) Trucks at km/h (0 mph) Passenger car SUV Van Articulated truck Heavy truck FIGURE Effect of Roughness on Repair and Maintenance Costs TRB 0 Annual Meeting

13 Zaabar, I. and Chatti, K. 0 R & M C ost The average repair and maintenance cost for roads with IRI>m/km (0 in/mile) is:.... FIGURE Road Surface Roughness Distribution in the U.S. A computer program was also developed to facilitate the use of the model. The program can estimate repair and maintenance costs at the project and network levels. For project level - analysis, the actual road profile should be used to account for the effect of roughness features. CASE STUDIES ( ) (, 000 m iles/year/car) = m illion cars ( $0.0 to $0.0 ) / m ile =. to. $ B illion/year Probability density function (%) 0% 0% 0% % 0%.% 0.%.%.%.%.%.%.%.%.% This section shows examples on how the VOC models can be used in practice. Three different examples are presented: () sensitivity analysis, () project level analysis, and () network level analysis. The unit costs were presented in TABLE. The software developed in this study was used to generate the results presented below. The VOC module is an engineering software application that allows one to calculate vehicle operating costs at the network and project levels. For network analysis, data for traffic, environmental and pavement conditions (IRI, MPD and pavement type) are the input to the module. For project analysis, one can import profiles in text format and the module will calculate the IRI. Entire analysis projects can be saved, which preserves user information and analysis inputs. After analyses have been performed, one can export a report of the results of any analyses. Example - Sensitivity Analysis In this example, the sensitivity of the total VOC to pavement conditions at, and km/h (, and 0 mph) is investigated. The pavement conditions considered are IRI and texture. TRB 0 Annual Meeting

14 Zaabar, I. and Chatti, K. Effect of Roughness on VOC The effect of pavement roughness (IRI) on VOC is estimated for all vehicle classes using the models developed in this study. TABLE presents examples of total cost expressed in cents ( ) per mile. The effect of roughness on VOC follows an exponential trend for all vehicle classes. The table was generated at C (. F), which is the average temperature in the U.S., with MPD of mm (0.0 in) and grade of 0%. TABLE Effect of Roughness on Vehicle Operating Costs Speed Vehicle Class Total Vehicle Operating Costs per Vehicle ( /mile) 0 or or or Medium car Van SUV Light Truck Heavy Truck Arti. Truck Medium car Van SUV Light Truck Heavy Truck Arti. Truck Medium car Van SUV Light Truck Heavy Truck Arti. Truck km = 0. mile; mm = 0.0 in; m/km =. in/mile; MPD= mm, Grade= 0%; Temperature = C (. F). Effect of Texture on VOC The effect of pavement surface texture (MPD) on VOC is investigated for all vehicle classes using the models developed in this study. TABLE presents examples of total cost expressed in cents ( ) per mile. The table was generated at C (. F), with IRI of m/km (. inch/mile) and grade of 0%. Discussion The combined effect of MPD and IRI can be predicted by multiplying the roughness and texture factors. For example, if one would like to estimate the total VOC for IRI = m/km (0 in/mile) and MPD = mm, for an articulated truck at km/h ( mph), divide. by 0. from TABLE, then multiply this ratio by. (TABLE ). The cost obtained for these conditions is cents/mile. TRB 0 Annual Meeting

15 Zaabar, I. and Chatti, K. 0 TABLE Effect of Texture on Vehicle Operating Costs Speed or mph or or 0 Vehicle Class Total Vehicle Operating Cost per vehicle ( /mile) Mean Profile Depth (mm).. Medium car Van... SUV.... Light Truck... Heavy Truck.0... Arti. Truck....0 Medium car.... Van SUV Light Truck..0.. Heavy Truck Arti. Truck Medium car.... Van.... SUV.. Light Truck.... Heavy Truck.... Arti. Truck..0.. km = 0. mile; mm = 0.0 in; m/km =. in/mile; IRI= m/km, Grade= 0%; Temperature = C (. F). Example - Project level analysis This example uses the M-E approach developed in this study to calculate the VOC costs (FC, TW and R&M) for a. km (. miles) long rigid pavement section on I- near Lansing, MI. The Average Daily Traffic (ADT) for this section is, in both directions, with 0% passenger cars, % commercial trucks, % heavy trucks, % SUV, % vans, % light trucks, and % buses (). The pavement condition data (raw profile and texture depth) were collected by MDOT using a Rapid Travel Profilometer and a Pavement Friction Tester. The grade was measured using a high precision GPS. FIGURE summarizes the distributions of its pavement conditions. The following procedure was followed to calculate VOC: - For R&M costs, the profile was input to the computer program developed as part of this study. The software calculated the accumulated damage in the suspension system, which was translated into R&M costs. - For FC and TW, the raw profile was divided into 0. km (0. mile) long subsections, and the IRI values were computed for each subsection. The other pavement conditions (grade, texture depth, and curvature) were input to the calibrated HDM models to estimate FC and TW. - The total costs were calculated according to the proportion of vehicle class mentioned above, and assuming average environmental conditions (Temperature = C (. F)). TRB 0 Annual Meeting

16 Zaabar, I. and Chatti, K. Number of subsections m/km =. in/mile 0 mm = 0.0 in MPD (mm) (a) IRI Distribution (b) Texture Distribution Number of Subsections Grade (%) (c) Grade Distribution FIGURE Pavement Conditions along I- Project Number of Subsections 0 FIGURE shows the costs for each susbsection (0. km or 0. mile) for the traffic distribution generated at km/h (0 mph) for trucks and buses and at km/h (0 mph) for passenger cars, vans and SUVs. Each point represents a subsection. To estimate the reduction in VOC from rehabilitating the I- project, a raw profile of a newly overlayed pavement with an average IRI of m/km (. in/mile) was simulated. It was assumed that the grade and texture distribution were not affected by the rehabilitation. FIGURE shows the reduction in VOC for each subsection. The total reduction in VOC from rehabilitating this project across. km (. miles) will be about $. Million per year. 0. Cost (Million $/year) Distance (km) Fuel Consumption Tire Wear Repair and Maintenance subsection = 0. km; km = 0. mile. FIGURE Costs per Year Induced along I- Project by Subsection TRB 0 Annual Meeting

17 Zaabar, I. and Chatti, K. 0 Reduction in VOC (Million $/year) 0.0 Smooth Section Rough Section Distance (km) km = 0. mile FIGURE Reduction in VOC from Rehabilitating the I- Pavement Project These costs could be considered in a life cycle cost analysis (LCCA). This detailed analysis would help identify the segments of the pavement section that would result in higher operating costs. These segments would be considered for early maintenance. Example - Network level analysis In this example the developed models are used to compare the influence of maintaining the entire network versus maintaining a proportion of it (e.g., 0 or 0%) for simulated pavement networks of urban interstate highways in different States. A roughness range of to m/km (. to in/mile) was assumed. FIGURE shows the assumed roughness distributions before and after rehabilitation. The distribution before rehabilitation was obtained by specifying a normal distribution with the desired IRI range. For the other two distributions, an IRI value of m/km (. in/mile) was assigned to rehabilitated sections. The remaining sections were then randomly assigned an IRI value from the original distribution. The vehicle kilometers traveled (VKT) for each State was estimated using Table VM- and VM- from Highway Statistics (). TABLE shows the speed limit for trucks and cars by State used in this example (). TABLE presents the estimated reduction in VOC resulting from rehabilitating 0% versus 0% of the network for each State. According to a study conducted by the Pennsylvania Transportation Institute (Kilareski et al., 0), % of the road network in the U.S are flat and straight (grade is 0% and superelevation is 0%). Therefore, at the network level, assuming a grade of 0% and a super-elevation of 0% are reasonable assumptions to calculate VOC savings. TRB 0 Annual Meeting

18 Zaabar, I. and Chatti, K. Probability Distribution (%) Before Rehabilitation After Rehabilitation of 0% of the Network After Rehabilitation of 0% of the Network FIGURE Assumed Roughness Distribution for Pavement Network TABLE Speed Limits Used in Analysis STATE Cars URBAN INTERSTATES Trucks Cars Trucks STATE Cars URBAN INTERSTATES Trucks Cars Alabama Montana Alaska Nebraska Arizona Nevada Arkansas New Hampshire California New Jersey Colorado New Mexico Connecticut New York Delaware North Carolina 0 0 Dist. of Columbia North Dakota Florida Ohio Georgia Oklahoma 0 0 Hawaii Oregon Idaho Pennsylvania Illinois Rhode Island Indiana South Carolina 0 0 Iowa South Dakota Kansas 0 0 Tennessee 0 0 Kentucky Texas 0 0 Louisiana 0 0 Utah Maine Vermont Maryland Virginia 0 0 Massachusetts Washington 0 0 Michigan 0 0 West Virginia Minnesota Wisconsin Mississippi 0 0 Wyoming 0 0 Missouri 0 0 U.S. Total Trucks TRB 0 Annual Meeting

19 Zaabar, I. and Chatti, K. TABLE Estimated Vehicle Operating Costs per Year for Urban Interstates Highways in the US STATE Vehicle Operating Costs per Year ($ Billions) Reduction in VOC per Year ($ Millions) STATE Vehicle Operating Costs per Year ($ Billions) Reduction in VOC per Year ($ Millions) Original 0% 0% 0% 0% Original 0% 0% 0% 0% Alabama Montana Alaska Nebraska Arizona Nevada..... Arkansas..... New Hampshire California New Jersey..... Colorado New Mexico Connecticut..... New York Delaware North Carolina Dist. of Columbia North Dakota Florida..... Ohio..... Georgia..... Oklahoma..... Hawaii Oregon Idaho Pennsylvania Illinois..... Rhode Island Indiana..... South Carolina Iowa South Dakota Kansas..... Tennessee Kentucky Texas..... Louisiana..... Utah Maine Vermont Maryland..... Virginia Massachusetts Washington Michigan..... West Virginia Minnesota Wisconsin..... Mississippi..... Wyoming Missouri..... U.S. Total scenario where 0% of the network is rehabilitated scenario where 0% of the network is rehabilitated CONCLUSION The objective of this study was to investigate the effect of pavement conditions on Vehicle Operating Costs (VOC) including fuel consumption, tire wear and repair and maintenance. The research does not include the effect of pavement conditions on changes in travel time, nor does it consider the safety-related, environmental, or other implications of pavement conditions. The paper showed summary results and project and network level examples on how to use the TRB 0 Annual Meeting

20 Zaabar, I. and Chatti, K calibrated VOC models in pavement management decisions. Some specific conclusions that were arrived at include the following: This study demonstrated that vehicle operating costs increase with pavement roughness across all classes of vehicles and types of pavements investigated. The most important cost components affected by roughness are fuel consumption followed by repair and maintenance, then tire wear. For fuel consumption, the most important factor is surface roughness (IRI). An increase in IRI of m/km (. in/mile) will increase the fuel consumption of passenger cars by % to % irrespective of speed. For heavy trucks, this increase is % to % at highway speed ( km/h or 0 mph) and % to % at low speed ( km/h or mph). Surface texture (MPD) and pavement type have no effect on fuel consumption for all vehicle classes with the exception of heavy trucks. An increase in MPD of mm (0.0 in.) will increase fuel consumption by about.% at km/h ( mph) and about % at km/h ( mph). Heavy trucks driven over AC pavements will consume about % more fuel than over PCC pavement at km/h in summer conditions. The effect of pavement type was statistically not significant at higher speeds. No data was available for heavy trucks in winter. For repair and maintenance (R&M), there is no effect of roughness up to IRI of m/km (0 in/mile). Beyond this range, an increase in IRI up to m/km ( in/mile) will increase R&M cost by % for passenger cars and heavy trucks. At IRI of m/km ( in/mile), this increase is up to 0% for passenger cars and 0% for heavy trucks. For tire wear, only the effect of roughness was considered. An increase in IRI of m/km (. in/mile) will increase the tire wear of passenger cars and heavy trucks by % at km/h ( mph). It is recommended to use the proposed calibrated models and the corresponding computer program to estimate vehicle operating costs at the project and network levels. For project level analysis, the actual road profile should be used to account for the effect of roughness features. Finally, it should be noted that growing demand for fuel efficient vehicles has accelerated the research and development (R&D) efforts to meet this demand. Therefore, new engine and combustion technologies, alternative fuels, vehicle design and maintenance, and tire technologies will affect vehicle operating costs in the future. This means that the recommended mechanisticempirical models would have to be calibrated to address emerging technologies. ACKNOWLEDGMENTS This research was conducted as part of the NCHRP project -. The authors would like to acknowledge the financial support of the National Cooperative Highway Research Program (NCHRP) of the National Academies, the input of the technical panel and the TRB senior program manager, Dr. Amir Hanna. The authors also would like to thank the technical support from MDOT and Texas DOT for providing the repair and maintenance data of their fleet. TRB 0 Annual Meeting

21 Zaabar, I. and Chatti, K REFERENCES. Chatti, K. and Zaabar, I., Estimating the Effects of Pavement Conditions on Vehicle Operation Costs, TRB s National Cooperative Highway Research Program (NCHRP) Report 0, 0.. Bennett, C. R., and Greenwood, I. D., Volume : Modeling Road User and Environmental Effects in HDM-, Version.0, International Study of Highway Development and Management Tools (ISOHDM), World Road Association (PIARC), ISBN: -00--, 00.. Zaabar, I., Effect of Pavement Condition on Vehicle Operating Costs Including: Fuel Consumption, Vehicle Durability and Damage to Transported Goods, Ph.D. Dissertation: Department of Civil and Environmental Engineering, Michigan State University, 0.. Zaabar, I. and Chatti, K., Calibration of HDM Models for Estimating the Effect of Pavement Roughness on Fuel Consumption for U.S Conditions, Transportation Research Record: Journal of the Transportation Research Board, Transportation Research Board of the National Academies, -, pp -, 0.. Zaabar, I. and Chatti, K., A Field Investigation of the Effect of Pavement Type on Fuel Consumption, Proceedings of the First T&DI Congress: Integrated Transportation and Development for a Better Tomorrow, 0.. Zaniewski, J. P., Butler, B. C., Cunningham, G., Elkins, G. E., Paggi, M. S., and Machemehl, R. (). "Vehicle Operating Costs, Fuel Consumption, and Pavement Type and Condition Factors." FHWA-PL--00, Texas Research and Development Foundation, Austin, Texas.. Zaabar, I. and Chatti, K., A New Mechanistic-Empirical Approach For Estimating The Effect Of Roughness On Vehicle Durability, Transportation Research Record: Journal of the Transportation Research Board, Transportation Research Board of the National Academies,, pp 0-, 0.. Michigan Department of Transportation, "00 Average Daily Traffic (ADT) Maps", (Accessed November 0). Federal Highway Administration, "Highway Statistics, US department of Transportation." Governors Highway Safety Association, (Accessed May 0) TRB 0 Annual Meeting